Loss of NSX Edge RTEP connectivity between two data centers

Products

• VMware NSX

Issue/Introduction

Cross-Site Remote Tunnel Endpoint (RTEP) loss of connectivity between two data centers preventa multi-site overlay communication.

NSX UI shows RTEP with DOWN state 

Environment

VMware NSX 

Cause

Direct inspection via the Edge CLI command

'get bgp neighbor summary'

confirms that the BGP peering session does not in the Established state.

Packet captures on the ESXi host vmnic show outbound BGP traffic (TCP port 179) directed toward the remote site TEP, but no return traffic (ACK) is received.

The underlying BGP peering between Edge Nodes (en01) across data centers is failing to establish because BGP control plane traffic (TCP 179) is being dropped or blocked by an intermediate network firewall.

RTEP establishment in a multi-site NSX architecture requires BGP to exchange TEP reachability information. Evidence from CLI diagnostics shows the BGP session is stuck in Active or Idle, and packet captures confirm a lack of ingress BGP packets, indicating a network-level blockage rather than a protocol mismatch or configuration error.

Resolution

 The network security team to permit bidirectional traffic on TCP Port 179 (BGP) between the RTEP IP addresses of the Edge Nodes in both datacenters.

Opening TCP port 179 enables the BGP three-way handshake and subsequent peering. Once the BGP session is Established, the Edge Nodes can exchange prefix information, allowing the RTEP tunnels to transition to an UP state for data plane encapsulation.